European robotics companies are increasingly adopting NVIDIA’s Isaac platform, Omniverse simulation tools, and the newly introduced Halos safety architecture to support the development of autonomous robotic systems designed to operate in physical environments. This development was discussed during the GTC conference held in Paris in conjunction with VivaTech, where several robotics firms showcased systems built on NVIDIA’s technologies.
Participating companies included Agile Robots, Extend Robotics, Humanoid, idealworks, Neura Robotics, SICK, Universal Robots, Vorwerk, and Wandelbots. These organizations are incorporating various components of NVIDIA’s AI stack, which now includes updated versions of its simulation and training tools. Among the updates is Isaac GR00T N1.5, a humanoid robot foundation model that offers improved reasoning and task instruction-following capabilities. This model is now available through Hugging Face. Additionally, Isaac Sim 5.0 and Isaac Lab 2.2 have been made available as open-source tools in developer preview on GitHub, optimized for NVIDIA RTX PRO 6000 workstations.
NVIDIA also introduced Halos, a robotics safety architecture that combines hardware, AI models, software tools, inspection mechanisms, and certification services. Halos is now part of an ANAB-accredited oversight program, with companies such as Boston Dynamics and SICK engaging in certification activities through the Halos Inspection Lab.
Several firms have already begun integrating these tools into their operations. Agile Robots is using Isaac Lab and the GR00T N1 model to train dual-arm manipulators on Jetson hardware. idealworks is implementing Omniverse’s mega blueprint for fleet simulation and adapting it to humanoid workflows. Neura Robotics is employing GR00T-Mimic to enhance its MiPA service robot, linking it to SAP’s Joule agents and validating outcomes through Omniverse simulations. Vorwerk is developing custom humanoid systems using Isaac Lab and deploying them via Jetson AGX, Orin, or Thor computing modules. Humanoid has reported a six-week reduction in development time by using NVIDIA DGX systems for training vision-language-action models. Universal Robots has announced the UR15 cobot, which uses an AI accelerator developed with Isaac CUDA libraries and Jetson AGX Orin. SICK has incorporated its 2D and 3D lidars and cameras into Isaac Sim to model and test perception-based systems within the Omniverse environment.
